Previously, we checked that the event queue had an OMX_EventCmdComplete
item, but we didn't make sure that it was for the previously issued
OMX_CommandStateSet. In many cases, it was from a OMX_CommandFlush,
which made the code proceed with other deinitialization. If the decoder
hadn't actually transitioned to idle state yet, the buffers weren't
actually ever freed (in the state == OMX_StateIdle block), which lead
to crashes when the handle was freed at the end.

This fixes crashes when finishing playback of wmv3 videos on Galaxy S3.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This one doesn't seem to work for either VC1 or WMV3 at the moment
(on Android 4.1.2 on Galaxy S3), while OMX.SEC.vc1.dec works (for
both codecs).
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Some OMX decoders that can handle WMV3 (such as the broadcom one
in raspberry pi) can handle the WMV3 extradata both in the original
form as in ASF, and in this format, while the one in Samsung Galaxy
S3 requires it to be in this format (as the OMX IL 1.2 specs specify).

This makes HW-accelerated WMV3 decoding work on Galaxy S3.

The code can be split out to a reusable function if/when the same
needs to be done elsewhere.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Rafaël Carré <funman@videolan.org>

Signed-off-by: Rafaël Carré <funman@videolan.org>

Signed-off-by: Rafaël Carré <funman@videolan.org>

Signed-off-by: Rafaël Carré <funman@videolan.org>

These are required by the spec.  We just set them to true, to match the
fact that the buttons on the main interface are never disabled.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

There was a lot of repeated code, some of which properly checked return
values and some of which didn't.  This is all replaced and folded into
the GetProperty function of each interface.

A side-effect is that it should be more efficient (we only check the
interface argument once instead of once for each property, and there
should be a smaller overall code size).
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Pull out code to marshal the various properties into separate functions
(and make sure the return type of those functions is int).  This is in
preparation for adding org.freedesktop.DBus.Properties.GetAll support.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Re-use the AddProperty method (used in GetAllProperties) and use macros
to reduce repetition of code.  Now all the calls should be
error-checked.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

 * Add missing call to vlc_object_release() in MarshalMetadata
 * Separate code paths with empty lines
 * Fix bracket position in if blocks

VLC was getting kicked from the D-Bus when a track was stopped, because
it was generating invalid data on the wire when sending the
PropertiesChanged signal for the Metadata property.

The issue was that if there was now no current track, GetInputMeta would
never be called and the the variant would never be populated with the
"a{sv}" structure that the call to dbus_message_iter_open_container
claimed it would be.

We now share the code that GetProperties used, which dealt with this
correctly, although now both use CurrentInput (which is what the signal
previously used) instead of CurrentPlayingItem (which is what
GetProperties previously used).
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Rate should never be 0.0 (according to the spec), so we pick a sensible
default when we do not have a current track: 1.0.

Also, when the current item changes, the Rate may also change.  So add
it to the PropertiesChanged signal just to be sure.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Several of these entries, such as xesam:artist, should be lists of
strings, not strings.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

Previously, Fullscreen could be requested via
org.freedesktop.DBus.Properties.Get, but would not be returned by
org.freedesktop.DBus.Properties.GetAll.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

This is in preparation for implementing GetAllProperties for the player
and tracklist interfaces.
Signed-off-by: Mirsal Ennaime <mirsal@videolan.org>

This is the right place to make sure an art file that was not available at
input launch time gets a chance to be displayed as soon as it becomes available.

disc.svg, network.svg derivative work from Gnome's elementary icons (GPL)

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

When MP4_ReadBoxContainerChildren() returns, one expects the position in
the stream to be that of the next box after the last parsed box.

Without this very patch, if the last parsed box is an "unknown" one,
the position in the stream would be at the begining of that box,
preventing playback of latest Anevia's MS Smooth Streaming live streams,
for example.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Since non-ASCII or non-printable characters are not supposed to occur
in URL, it does not matter much how they are dealt with.

The caller of decode_URI() can anyway not assume anything about the
decoded string. In particular, decode_URI() does not validate UTF-8
sequences in any case.

Pointed-out-by: Mario Speiß <1034-135@online.de>

When changing a skin theme on the fly, an input may already be running
 and the art file is expected to display in the new theme.

This patch takes into account a possible art file already kwown
at initialisation of a new image control with art enabled.

Art display used to rely on an event issued by vlc core which has been renamed into "activity" lately. The issue was that it seems no longer fired when a new
 input is launched.

A relative jump of +1 is the same than 0, we need +2 to skip the next
instruction.
Use named label so it is more clear than +2.

Read the InstallDir regkey, "" points to vlc.exe